1726
J. Phys. Chem. A 2006, 110, 1726-1734
Characterizing the Fluorescence Intermittency and Photobleaching Kinetics of Dye Molecules Immobilized on a Glass Surface Edwin K. L. Yeow,* Sergey M. Melnikov, Toby D. M. Bell, Frans C. De Schryver, and Johan Hofkens* Department of Chemistry, Katholieke UniVersiteit LeuVen, Celestijnenlaan 200F, B-3001 HeVerlee, Belgium ReceiVed: September 27, 2005; In Final Form: NoVember 29, 2005
The blinking behavior of single Atto565 molecules on a glass surface is studied under air or nitrogen atmospheres using confocal microscopy. The broad distributions for both on- and off-time durations obey power law kinetics that are rationalized using a charge tunneling model. In this case, a charge is transferred from the Atto565 molecule to localized states found on the glass surface. Subsequent charge recombination by back charge tunneling from trap to Atto565 cation (i.e., dark state) restores the fluorescence. The off-time distribution is independent of excitation intensity (I), whereas the on-time distribution exhibits a power law exponent that varies with I. Two pathways have been identified to lead to the formation of the radical dark state. The first involves direct charge tunneling from the excited singlet S1 state to charge traps in the surrounding matrix, and the second requires charge ejection from the triplet T1 state after intersystem crossing from S1. Monte Carlo simulation studies complement the two-pathway model. Photobleaching curves of both single and ensemble molecules do not exhibit monoexponential decays suggesting complex bleaching dynamics arising from triplet and radical states.
1. Introduction Photoblinking and photobleaching are important characteristics of single molecules. Photoblinking refers to the temporary disappearance of emitted light when molecules undergo reversible transitions between “on” and “off” states, whereas photobleaching is an irreversible process where bleached products that do not absorb or emit at the excitation and emission wavelengths, respectively, are formed from the molecule. In many applications of single molecule spectroscopy there is a pertinent need to increase the photostability of fluorescent probes and to avoid interruptions during light emission which may otherwise interfere with the actual processes under investigation (e.g., protein folding,1 enzyme catalysis,2 electron transfer3). To increase the efficacy of fluorescent probes, it is important to first understand the dynamics behind photobleaching and photoblinking. Even though many studies have been dedicated to unraveling the mechanisms responsible for these two processes, there still remain controversies surrounding the nature of the dark states, and the pathways leading to their formation.4-11 Fluorescence intermittency of single molecules has previously been explained in terms of triplet state dynamics,12,13 molecular reorientation,11 spectral diffusion,15 conformational changes,8,9 and intramolecular electron transfer.3,14 More recently, on/off events following power law distributions for an ensemble of single molecules have been reported. Haase et al. observed longlived dark states that obey power law distributions for perylenemonoimide chromophores embedded in poly(methyl methacrylate) film,16 and Schuster et al. reported similar power law blinking behavior for various dye molecules (e.g., rhodamine 6G, terrylene) dispersed on top of glass surfaces and polymer films.17,18 Hoogenboom et al. reported a power law distributed dark state for a trimer molecule consisting of three rigidly linked * Corresponding authors. E-mail:
[email protected] (E.K.L.Y.),
[email protected] (J.H.). Fax: (+32) 16327989.
tetraphenoxyperylene diimide chromophores.19 A model based on the formation of (dark) radical cations arising from electron transfer between chromophores and self-trapped states found in the host matrix has been used to understand the sensitivity of power law kinetics to environmental polarity. The assignment of long-lived dark states responsible for photoblinking to radicals was further supported by Zondervan and co-workers’ study in which radical anions of rhodamine 6G (Rh6G), suggested to be formed via electron transfer from poly(vinyl alcohol) to Rh6G upon light irradiation, were observed in an electron-spinresonance experiment.20 A new range of rhodamine-type molecules called Atto dyes has recently been demonstrated to be potential labels for use in fluorescent imaging.21 In addition, the absence of cis-trans isomerization in these dyes enables one to neglect conformational changes as a source of photoblinking. We have chosen therefore to characterize the fluorescence intermittency and photobleaching kinetics of Atto565 (structure given in Chart 1) molecules immobilized on glass surface. Due to the long triplet T1 lifetime compared to the singlet S1, the T1 state has often been implicated in playing an active role in photoblinking and photobleaching processes.11,12,22 To understand the importance of T1 in affecting fluorescence intermittency, the T1 state of Atto565 was first characterized using fluorescence correlation spectroscopy. Confocal microscopy was then used to investigate the photoblinking and photobleaching kinetics where power law kinetics for the photoblinking events are observed for Atto565. There remains ambiguity concerning the pathways leading to the formation of the radical dark states.16-20 It has often been proposed that charge transfer occurs mainly from the triplet T1 state rather than from the singlet S1 state.20 However, detailed analysis of the possible pathways leading to the long-lived dark state of Atto565 in this work suggests the importance of charge transfer from the singlet S1 state to the surrounding traps.
10.1021/jp055496r CCC: $33.50 © 2006 American Chemical Society Published on Web 01/06/2006
Photobleaching Kinetics of Dye Molecules
J. Phys. Chem. A, Vol. 110, No. 5, 2006 1727
CHART 1: Chemical Structure of Atto565
2. Experimental Section In the fluorescence correlation spectroscopy (FCS) experiment, ca. 10-9 M aqueous solution of Atto565 (Atto-Tec) was excited at 543 nm using a continuous wave (cw) HeNe laser (Melles Griot). The excitation beam was directed into an inverted fluorescence microscope (Olympus IX 70), and focused onto the sample through an objective lens (Zeiss, 100x, N.A. 1.3, oil immersion). The fluorescence light was collected by the same objective and refocused through a pinhole (100 µm diameter). Unwanted scattered laser light was removed from the fluorescence signal using a notch filter, and the signal was then detected by an avalanche photodiode APD (SPCM 15, EG & G). The autocorrelation function was calculated from a digital correlator (ALV-5000, ALV GmbH). The experimental setup used to record emission intensity time traces of single molecules dispersed onto glass surfaces is described in detail elsewhere.23 Basically, an inverted microscope (Oympus IX 70) equipped with a scanning stage was used to detect single molecules. Time-resolved fluorescence decay curves were recorded by the time-correlated single photon counting technique (TCSPC) using a 543 nm pulsed laser light (8.18 MHz, 1.2 ps fwhm) excitation from a frequency doubled OPO (Spectra Physics) pumped by a mode-locked, regeneratively amplified Ti:Sapphire laser (Spectra Physics). The excitation light was focused through an oil immersion objective lens (Olympus, 100x, N.A. 1.4) which was also used to collect the fluorescence. A 50%/50% beam splitter divided the fluorescence signal into an APD for TCSPC measurements and into a CCD camera (LN/CCD-512SB, Princeton Instruments) coupled to a 150-mm polychromator (SpectraPro 150, Acton Research Coorp.) for fluorescence spectra measurements. Samples for single molecule experiments were prepared by spin-casting solutions of Atto565 in methanol (∼10-10 M) onto thoroughly cleaned borosilicate glass coverslips (Menzel-Glaser). The ensemble bleaching experiment was done by depositing a higher concentration of dye molecules (ca. 10-6 to 10-7 M) onto cleaned glass coverslips. At this concentration, aggregation and self-energy-transfer have been shown to be minimal.24,25 The experimental setup is identical to the one for FCS and the sample was excited using a cw HeNe laser. Both single molecule and ensemble measurements were performed in air and nitrogen atmospheres. The latter was achieved by gently blowing nitrogen through the sample holder during the measurement. 3. Results and Discussion 3.1. Fluorescence Correlation Spectroscopy. Figure 1 shows the fluorescence autocorrelation curve obtained for Atto565 in aerated water using an excitation intensity (I) of 11 kW/cm2. The autocorrelation curve was fitted to the function:26
Figure 1. Autocorrelation curve for Atto565 in air atmosphere. Excitation wavelength was 543 nm.
G(t) )
[
]
1 1 - F + F exp(-t/τtri) × Neff 1-F
[(
1+
t τdiff
)(
1+
t ω τdiff 2
)]
1/2 -1
(1)
where Neff denotes the average number of molecules residing in the effective probe volume, F is the average fraction of molecules in the triplet state, ω describes the shape of the ellipsoidal detection volume (ω ) zo/ro, where zo and ro are the axial and lateral radii, respectively), and τtri and τdiff are the triplet and diffusion relaxation time, respectively. The fitting analysis yielded values of 5.4 × 10-6 s and 0.0685 for τtri and F, respectively. When a uniform excitation profile within the probe volume is assumed, τtri and F are given by27
kexkisc 1 ) ktri + τtri ks + kex F)
kexkisc kex(ktri + kisc) + ksktri
(2)
(3)
where ks is the deexcitation rate from the excited singlet state (S1) including fluorescence and internal conversion (2.94 × 108 s-1), kisc is the S1-T1 intersystem crossing rate, and ktri is the decay rate from the triplet T1 state. The excitation rate, kex ()6.84 × 106 s-1), was calculated from kex ) σNI, where σ is the absorption cross section of Atto565 at 543 nm (2.32 × 10-16 cm2), and N is the number of photons in 1 J at 543 nm (2.73 × 1018 photons). From eqs 2 and 3, we obtained kisc ) 5.58 × 105 s-1 and ktri ) 1.73 × 105 s-1 for Atto565 in aerated water. These values are not significantly different from those determined for other rhodamine dyes such as rhodamine 6G (Rh6G) (kisc ) 1.1 × 106 s-1 and ktri ) 5.0 × 105 s-1 in water) and rhodamine 610 (kisc ) 2.0 × 105 s-1 and ktri ) 5.0 × 105 s-1 in ethylene glycol).27,28 3.2. Fluorescence Intensity Fluctuations. Typical emission intensity time traces of single Atto565 molecules on a glass surface under ambient conditions are shown in Figure 2. Figure 2a shows rapid emission intensity fluctuations with several long off periods. Similar photoblinking events with a duration lasting milliseconds to a few seconds are observed for other Atto565 molecules on the same glass surface. In Figure 2b, the molecule displays nearly constant emission intensity before undergoing irreversible photobleaching at around 105 s. Furthermore, a 330 ms off-time period is observed at about 28.8 s (see Figure 2c). Several explanations have been provided in the past to explain fluorescence intermittency, and the nature of long-lived dark
1728 J. Phys. Chem. A, Vol. 110, No. 5, 2006
Yeow et al.
Figure 2. (a) Fluorescence intensity time trace of a single Atto565 molecule on glass surface with I ) 568 W/cm2 and in air atmosphere. (b) Fluorescence intensity time trace of another single Atto565 molecule on glass surface with I ) 568 W/cm2 and in air atmosphere. (c) Fluorescence intensity trace of (b) from 28.4 to 29.4 s magnified to show clearly an off-time of 330 ms at around 28.8 s. (d) The fluorescence intensity trace of (b) from 28.4 to 29.4 s with a smaller bin time of 1 ms.
Figure 3. (a) Position of the maximum emission intensity wavelength (λmax) vs time for the emission spectra recorded with 5 s integration time for the emission intensity time trace shown in the inset. (b) The emission spectra after 10 s (point x in Figure 3a) and 25 s (point y in Figure 3a).
states in dye molecules, namely, (1) intersystem crossing from an excited singlet state to a triplet (dark) state,12 (2) rotation of the molecular dipole moment giving rise to variations in the amount of light absorbed and emitted,11 and (3) spectral diffusion causing the absorption transition to shift in to and out of resonance with the excitation wavelength.8,9,15 Given that the triplet lifetime of Atto565 in air (i.e., ∼6 µs) is much shorter than the observed long off-times discussed above, the triplet state is not responsible for photoblinking events occurring in this time regime (i.e., >1 ms). Furthermore, it has been demonstrated using fluorescence polarization experiments that rhodamine dye molecules tend to be rigidly immobilized on glass surfaces, thus hindering any molecular reorientation.15,29 Rotational motion therefore plays an insignificant role in effecting fluorescence intensity fluctuations, especially because circular polarized light was used in all the single molecule measurements reported herein. The influence of spectral diffusion is most pronounced at low temperature, where, due to the narrow absorption bandwidth, a small spectral shift results in a substantial change in the absorption cross section at the laser excitation frequency. This effect is drastically reduced at room temperature where a large shift in the broad absorption band is needed to account for changes in the emission intensity. Figure 3a shows the distribution of the maximum emission intensity wavelength (λmax) as a
Figure 4. Off-time distributions for (a) 88 Atto565 molecules at 1136 W/cm2, (b) 86 molecules at 568 W/cm2, and (c) 73 molecules at 284 W/cm2 in air atmosphere. The linear log-log plots indicate power law distribution, and the solid lines are lines of best fit with power law exponents moff ) 1.8 (a), 2.0 (b) and 1.9 (c).
function of time for a single Atto565 molecule (emission intensity time trace is given in the inset of Figure 3a). The emission spectra measured with 5 s integration time after 10 s (point x in Figure 3a) and 25 s (point y in Figure 3a) are displayed in Figure 3b. From Figure 3a, we note that λmax varies between 571 and 583 nm. Assuming a Stokes shift of 870 cm-1 (from ensemble measurements), the absorption cross sections of the molecule at points x and y when excited at 543 nm differ by ca.. 1.2-fold which does not account for the ca. 5.3-fold reduction in the integrated fluorescence intensity (Figure 3b). Spectral diffusion is not, therefore, sufficient to describe completely the photoblinking events observed for Atto565 at room temperature, and other mechanisms must be sought to rationalize the long off-times. 3.3. On- and Off-Time Statistics. To characterize the fluorescence intermittency of Atto565 molecules, both on- and off-time duration histograms were constructed from emission time traces with 1 or 5 ms bin times. The number of counts associated with the observed detector background level after each molecule has undergone photobleaching is chosen to be the threshold value. The algorithm for creating on- and off- time histograms over large time range has been described in details elsewhere.8,30,31 An insufficient number of on/off events per molecule renders the statistical analysis based on a single molecule unreliable, and the combined distribution of on/off events for 70-90 molecules is used instead. A small fraction of the molecules studied (
